Group 29 Presentation

Introduction

  • CD molecules are membrane proteins with diverse functions and distributions across immune cell types.

  • Their expression patterns help distinguish cell lineages and reveal functional relationships

Aim:

  • How do the expression of CD markers on lymphocyte subsets change during maturation?

  • How are fluorescence intensity, variability, and positivity (MedQb, CVQb, PEpos) related across CD markers in lymphocyte subsets?

DOI: 10.5772/intechopen.81568

Materials and Methods

  • Data from: Frontiers in Immunology, “B Cell Biology,” vol. 10, Oct. 23, 2019. doi: 10.3389/fimmu.2019.02434

  • Can be downloaded from their shiny app: http://bioinformin.cesnet.cz/CDmaps/

  • The data set contains:

    • 28340 observations of 8 variables

    • 114 unique CDs and 38 unique cell types

Sample of data_aug:

# A tibble: 5 × 8
  tissue CD          lineage     cell_type hierarchy  CVQb  MedQb  PEpos
  <chr>  <chr>       <chr>       <chr>         <dbl> <dbl>  <dbl>  <dbl>
1 blood  CD28        B cells     BnatEff           3 819.   112.   0.644
2 blood  CD11b       T cells     Tgd               3 330    118.  11    
3 tonsil CD55        B cells     BnaiveTo          3  97.6 3301.  83.9  
4 blood  CD4_RPA-T4  CD8 T cells TCD8TEMRA         4 264.    31.2  1    
5 tonsil CD4_MEM-241 B cells     PC                3 114.  1281.   1.29 

Analysis 1

  • MedQb:
    • Three distinct clusters
    • Thymocytes cluster with T-cells from blood
    • B-cells form a tonsil cluster and a blood cluster
  • PEpos:
    • Four distinct clusters
    • Thymocytes and blood T cells are seperated in two clusters
    • B-cells from blood and tonsil seperate in PC2, but very similar in PC1

Analysis 2

  • Wide variations in CD marker distribution

  • Tissue-related clusters are common

  • Some markers are universally expressed (e.g. CD45), with others are lineage-specific

Analysis 3

CVQb:

  • Blue lines show a negative correlation → higher CVQb = lower MedQb

  • Tonsil B cells: line is flat → almost no relationship

PEpos:

  • Pink lines show a positive correlation → higher PEpos = higher MedQb

Analysis 4

  • How do CD marker expression change during maturation of each lineage?

    • Applying a linear model with naive cell as reference

  • B cells:

    • Had the most significant markers, showing stronger activation changes

    • Tonsil B cells are more activated (CD69, CD80), while blood B cells mature gradually (CD11a, CD80)

  • CD4 and CD8 T cells:

    • Follow a similar pattern

Analysis 5

  • CD4 and CD8 T cells go through parallel stages

  • How do the CD markers differ between CD4 and CD8 T cells for each stage?

    • As expected CD4 and CD8 are significant different for all pairs
    • CD59 is significant for all stages except TEMRA with a higher log(MedQb) for CD8
    • SP1am and Naive have the lowest number of significant CDs → CD4 and CD8 more similar in these stages

Discussion

Why did we choose the linear model to assess significant difference for CDs?

  • Simple method to compare each subset to the naive cell

  • An ANOVA could for example also have been used for pairwise comparison

Problems with missing values in the wide-format data set for PCA

  • Number of experiments for each CD differed → summarized the experiments by the mean

  • Not every CD marker was measured across all cell types → replaced the missing value with the median for that specific CD

  • Limits the variation in the data set, but necessary to avoid dropping observations